Identification of surrogate endpoints in patients with locoregionally advanced nasopharyngeal carcinoma receiving neoadjuvant chemotherapy plus concurrent chemoradiotherapy versus concurrent chemoradiotherapy alone

Table 3 Testing of Prentice’s criteria three and four for the potential surrogate endpoints for 5-year OS^a

Prentice’s criteria	HR (95 % CI)^c	P-value*
Prentice’s criterion three: Is the surrogate endpoint prognostic for 5-year OS?
Failure or death from any cause ≤ 3 years^b	39.0 (16.5–92.3)	<0.001
Failure at any site ≤ 3 years^b	14.9 (7.9–28.1)	<0.001
Distant metastasis ≤ 3 years^b	18.1 (9.7–33.8)	<0.001
Prentice’s criterion four: Is treatment still prognostic for 5-year OS in patients stratified by the surrogate?
Patients with failure or death from any cause ≤ 3 years
Yes	NS	>0.10
No	NS	>0.10
Patients with failure at any site ≤ 3 years
Yes	NS	>0.10
No	NS	>0.10
Patients with distant metastasis ≤ 3 years
Yes	NS	>0.10
No	NS	>0.10

Abbreviations: CI confidence interval; HR hazard ratio; NS not significant; OS overall survival
^aAs treatment was not prognostic for any surrogates measured at 2 years and locoregional failure at 3 years, Prentice’s criterion three and four were not tested for these endpoints
^bThe reference groups were no events ≤ 3 years
^cHazard ratios were adjusted for the following parameters using a Cox proportional hazards model by backward elimination: age (>50 vs. ≤ 50 years), gender (male vs. female), T category (T3–4 vs. T1–2), N category (N1–3 vs. N0), surrogates (with vs. without events ≤ 3 years) for Prentice’s criterion three only, and treatment arm (neoadjuvant chemotherapy + concurrent chemoradiotherapy vs. concurrent chemoradiotherapy) for Prentice’s criterion four only
*P-values were calculated using the adjusted Cox proportional hazards model

ISSN: 1471-2407